Home // INTELLI 2018, The Seventh International Conference on Intelligent Systems and Applications // View article
Authors:
Joao Reis
Gil Gonçalves
Keywords: Process Modeling; Process Parameter Optimization; Artificial Neural Networks; Smart Manufacturing; Machine Learning.
Abstract:
One of the main challenges towards a smart factory is the automation of processes and inclusion of personnel experience in those systems. One of these challenges is related to advances in artificial intelligence that have already been proven to be effective in solving real world problems in the last decade. The problem addressed in this paper is finding the most suitable machine parameters of a laser seam welding process. Once new quality requirements are defined by the customer, normally, a machine calibration phase is required in order to find the proper parameters that yield the desired quality of the product. To address this problem, first a modeling phase was performed to create a suitable model using Artificial Neural Networks (ANNs) that map process parameters onto the observed product quality, and second, the Basin-Hopping search algorithm was used to find the machine parameters needed to achieve a target quality. In order to demonstrate the robustness of the presented approach, three datasets were used that represent three different pairs of materials used for welding in the same machine. The results demonstrate that ANNs are a flexible and robust technique to be used in industry for process modeling and the calibration phase can be minimized.
Pages: 30 to 35
Copyright: Copyright (c) IARIA, 2018
Publication date: June 24, 2018
Published in: conference
ISSN: 2308-4065
ISBN: 978-1-61208-646-0
Location: Venice, Italy
Dates: from June 24, 2018 to June 28, 2018